EP1749395B1 - Method for authorizing a telephone transaction and switching node - Google Patents

Description

The invention relates to the authorization of a voice-aware telephonic transaction, and more particularly to a method of authorizing user access to information in which the user is prompted by an interactive voice device to deliver a voice sample, compare the voice sample to a comparison data set, and the access to the desired information by the user is enabled by the interactive voice device in dependence on the comparison result, as well as to a corresponding switching node.

The telephone authentication by means of voice recognition is becoming more and more important in voice dialog applications in the public telephone network. Before a user can use the service (e.g., telephone banking) of a service provider, he must identify himself. For this purpose, the service provider has one or more speech patterns of the user. The user identifies himself in this case by entering his name or user / access number and then complaining a speech pattern. By comparison with the deposited speech pattern, the user is identified and can then control his transactions with a positive comparison result.

Since the analysis of the language (speaker identification or verification) is very complex, a dedicated system (called authentication system in the following) is usually used for this purpose. There are a number of products that realize this as a standalone solution and provide proprietary or voice XML interfaces.

Out EP 1 172 771 A method is known for authorizing a commercial transaction between a customer and a merchandise or service provider by an authorization service provider with authentication of the customer by comparing a biometric sample of the customer with a customer biometric template stored with the authorization service provider. The authorization process here comprises the steps of (a) establishing a telephone connection via a communication network between the provider and the authorization service provider via a router, the router assigning the virtual telephone number the real telephone number of the authorization service provider and transmits the telephone number of the provider and a code that transmits at the same time the router and the authorization service provider is calculated using the same algorithm; (b) the authorization service provider accepts the telephone connection when the provider is registered with the authorization service provider at the transmitted telephone number and the one transmitted (C) transmission of a customer identifier to the authorization service provider over the established telephone connection, (d) transmission of a biometric sample of the customer over the established telephone connection, (e) transmission of the height of the code (f) initiation of the transaction by the authorization service provider, if the match between the biometric sample and the customer's stored biometric sample is greater than a predetermined threshold , having.

Out US 5,806,040 For example, a method and apparatus for verifying a person's identity using a calling card, bank card or credit card is known. The system establishes a connection between the user of the card and a telephone network, whereby the user enters the card number. The card number is used to access a central database via the telephone network, whereby a voice recognition pattern corresponding to the entered card number is loaded. The voice recognition includes various samples of the user's voice in the pronunciation of selected words, phrases and / or numbers. Once the corresponding voice recognition pattern has been loaded, the system causes the user to speak one of the words, sentences and / or digits that are in the voice recognition pattern. Word, sentence and / or number spoken upon request is compared with the voice recognition pattern. If the spoken words match the pattern, it is assumed that the user is authorized and the card is released. If the spoken words do not match the pattern, the user may or may not be denied authorization depending on the use of the card and the financial scope of the card.

In speech dialogue systems of this type, it is necessary to provide interfaces and operate a considerable effort in the adaptation of systems to integrate the identification or speaker verification in their operations, especially if a telephone system to be retrofitted with an authentication system and / or the telephone system and the Authentication system come from different manufacturers.

Out US Pat. No. 6535582 B1 is a method for authorizing access to a database by a user in a telecommunications network, the telecommunications network having a user interface, an access control for the database, and an authentication device.

The object of the present invention is to specify a method for authorizing a telephone transaction, ie a method with which the authorization of access to data via a public telephone network is reliably guaranteed and at the same time the effort in terms of user networking, interactive voice dialogue system and Authentication device is simplified. In particular, a retrofitting of existing systems is to be made possible.

This object is achieved by the method according to claim 1 and by the switching node according to claim 9. Preferred embodiments of the invention are subject of the dependent claims.

The inventive method for authorizing access to a database by a user in a telecommunications network, the telecommunications network having a user interface, access control for the database, and authentication means interconnected by a switching node in the telecommunications network, comprising the steps of: causing the user by the access control to deliver a voice sample, comparison of the voice sample with a comparison data set associated with the user and stored in advance in a memory, by the authentication means, release or denial of access to the desired information by the user in response to the comparison result by the access control is characterized in that the comparison result is output from the authentication device via the switching node to the access control and the comparison result by the access control is communicated to the user via the switching node, with signals from the authentication device being hidden.

In particular, the telecommunications network is an ISDN network.

In a preferred embodiment of the method according to the invention, the B-channels of the ISDN network are interconnected such that only the respective reception time slot for signals from the authentication device is mixed with a transmission time slot for signals for the user's telephone in which no authentication signals were received.

The comparison result is preferably transmitted in the form of DTMF sequences between the authentication device and the access control.

In another preferred embodiment of the invention, the voice sample comprises at least one sequence of multiple syllables or words, wherein the user is repeatedly asked to pronounce a word and / or syllable, recording one voice sample at a time, comparing each of the multiple voice samples to a comparison data set , wherein a hit rate is determined, and the release or denial of access to the desired information by the user as a function of the determined hit rate by the interactive voice device takes place.

Speech filtering is most preferably used to eliminate telephone line noise.

In particular, the suppression of signals from the authentication device takes place from the signals from the access control for the user in the switching node.

In a preferred embodiment of the invention, the access control is an interactive voice system.

The corresponding switching node according to the invention for switching signals in a telecommunication network between a user telephone, an authentication device and a database for authorizing access to a database by a user in a telecommunication network is characterized in that those signals transmitted from the authentication device to the access control of the database from the signals from access control of the database to the user's phone are hidden.

In a preferred embodiment, the switching node is an ISDN switching node in which the B-channels are interconnected such that only the respective time slot for data from the authentication device is mixed with a transmission time slot for data for the user's telephone in which no authentication signals were received.

Most preferably, the switching node includes voice filtering to eliminate telephone line interference.

An advantage of the described solution is that there are no security problems, since apart from the voice channel, no further interface to the existing authentication system is needed. In particular, the possibility of intercepting the communication between the authentication device and the database is thereby also prevented, so that in principle there is no opportunity to record signals from the authentication device for release / access denial. Furthermore, in the method according to the invention and the corresponding switching node according to the invention, the actual authentication device can be retrofitted and set up at any location in the telecommunications network.

Further advantages, characteristics and features of the solution according to the invention will become apparent from the following description of preferred embodiments, wherein reference is made to the accompanying drawings.

Fig. 1 shows schematically a system for carrying out a transaction over a public telecommunication network according to the prior art.

Fig. 2 shows schematically a system according to the invention for carrying out a transaction via a public telecommunications network.

In the following, the implementation of a transaction over a public telecommunications network according to the prior art is described with reference to FIG. 1, in which an authentication is carried out in advance, i. the access authorization is checked by a clear identification of the user.

In Fig. 1, a user 1a is connected via his telephone 1 to a telecommunication network 3. Via the telecommunication network 3, he can basically access a database 2 also connected to the telecommunication network 3. In the event that personal and confidential data is stored in this database, the access to the data should, of course, remain restricted to the actually authorized group of persons. For this purpose, the database 2 is preceded by an access control 2a, which checks the authorization of the user, who receives a corresponding request, before accessing the data.

In the event that the authentication of the user via a voice sample can be made, the request of the user 1a is returned from the access control 2a to a switching node 4, from which it is in turn forwarded to an external authentication device 5. The switching node is a common node located inside or outside the network 3. If the telecommunications network is an ISDN network, then in the switching node 4 in each case a transmission time slot is mixed with a reception time slot, so that the interconnected participants can communicate with each other as via a fixed line. In other words, one participant hears all the signals generated by the other two participants, there is no filtering.

In the authentication device 5, the voice sample of the user 1a is compared with comparison data which has been pre-stored in a local memory (not shown) for that user. The correct assignment of the comparison data to the user can be done via an entered access or identification number, which in turn can be spoken or entered as a number. The result of the comparison is returned via the same lines 6 as in the transmission of the request to the access control 2a. From the access control 2a, the user then receives the release for access to the database 2, or it is informed that he has no authorization to access. The notification of the access control 2a by the authentication device 5 can be made via an API (Application Programming Interface) interface.

In particular, the access control 2a is an interactive voice system so that the user 1a can be prompted by the local staff to speak a randomly selected word or sentence, or a randomly selected number, etc., the spoken text then being written to the authentication device as described 5 is compared with previously stored comparison data.

One application is e.g. a cash transaction by a bank employee after a previous authentication. For example, the customer (caller) 1a calls his bank and is directed to a voice dialogue system. The voice dialogue system establishes a conference call with the customer and with the authentication system as participants. The further control of the authentication takes over the authentication system 5. In this authentication, for example, a predefined sentence is queried and compared with a pattern. The acknowledgment of the authentication to the speech dialogue system 2a is done via DTMF, i. the authentication system 5 sends defined DTMF sequences, with which the success or the failure of the authentication is reported.

In the above-described conference circuit with the user's telephone 1, the switching node 4, the database 2 and the authentication device 5, which are connected to each other via the network 3, there is a risk that an eavesdropper or the caller himself records the acknowledgment signals and they at a next Authentication procedure in order to outsmart the system and gain access to confidential data that would otherwise have been denied him.

In order to prevent such, fraudulently attempted bypassing of the system, a conference circuit is constructed according to the invention, in which the response of the authentication system is received only by the voice dialogue system, but not by the caller itself.

Referring to Fig. 2, a "remote authentication" or "remote authentication" as an embodiment of the invention will be described below.

The caller inputs data via the telephone receiver or the keypad of his telephone 1, by means of which he establishes a connection with the access control 2a to the database 2 becomes. This can be an access number, a secret number (PIN) or even a voice signal. These data are first forwarded via the telephone network 3 to the switching node 4. In the switching node 4, the data required for authentication is forwarded to the authentication device 5, that is, for example, both DTMF signals input via the keyboard of the user's telephone 1 and voice signals input via the handset of the user's telephone 1 were. However, only the voice signals can be transmitted from the caller to the authentication device 5. In the authentication device 5, at least some of the characteristic data is compared by the caller with a data record which had previously been stored for this caller.

From the access control 2a of the database 2, the user is asked to certain words, syllables, sentences, digits u. Like to repeat. These voice data are - if necessary digitized - sent directly or from the access control 2a via the switching node 4 to the authentication device 5. FIG. 2 shows an embodiment in which the data is sent directly from the user telephone 1 to the authentication device 5. In the authentication device 5, as explained above with reference to FIG. 1, the speech data is compared with previously stored pattern or comparison data.

The comparison result is communicated by the authentication device 5 via the switching node 4 of the access control 2a of the database 2, preferably via a standard telephone line 6 by means of a DTMF sequence. In the access control 2a of the database 2, the result is received and analyzed. Subsequently, the result of the access control of the database 2 is communicated to the user via the switching node 4 and the standard telephone line 6, via which it is connected to the switching node 4. However, in order to prevent the result from the authentication device 5 from being recorded, the message is not sent to the user in the form of the same DTMF signals with which the result was signaled by the authentication device 5 of the access control 2a of the database 2 but, for example, in the form of Speech signals - if the access control in the form of a "call center" - is staffed with people - or in the form of other DTMF signals, but which are formed after a new "alphabet", so that they are different from the DTMF sequences of the authentication device 5 differ.

In the following, the function of the switching node 4 in the inventive method will be described in more detail. The data flows via the line 6 in the user communication, access control 2a, authentication device 5 and switching node 4 form a closed path, namely a data flow 6a from the user's telephone 1 to the switching node 4, a data flow 6b from the switching node 4 to the authentication device 5, a data flow 6c from the authentication device 5 to the switching node 4, a data flow 6d from the switching node 4 to the database 2 or its upstream interactive speech dialogue system 2a, a data flow 6e from the database 2 or its upstream speech dialogue system 2a to the switching node 4 and a data flow 6f from the switching node 4 to the user's telephone 1.

In this case, voice and other signals are transmitted from the user telephone 1 to the authentication device 5 by the data flows 6a and 6b via the switching node 4. These data are indicated by the hatched arrow 7. The data, which are transmitted as data flows 6c and 6d from the authentication device 5 via the switching node 4 to the interactive voice dialogue system 2a, possibly containing. The voice signals of the caller and also signals indicating the result of the comparison of the voice signals of the caller with corresponding stored samples , The latter signals are preferably DTMF signals. The data are indicated in FIG. 2 by a second, black arrow 8. In the voice dialogue system 2a, the DTMF signals are filtered out by the authentication device 5, so that only "uncritical" signals remain, namely the message to the user 1a that he or she can access the database 2. These message signals are sent as data flows 6e and 6f via the switching node 4 to the user's telephone 1.

The filtering of the DTMF signals from the authentication device 5 from the signals is achieved by a corresponding interconnection of the B (signaling) channels, wherein only the respective reception time slot for data from the authentication device 5 in the switching node 4 with a transmission time slot for data for the User phone 1 is mixed, in which no authentication signals were received. If necessary. this means that no reception time slot for data is mixed by the authentication device 5 in the switching node 4 with a transmission time slot for data for the user telephone 1. Thus, the answer of the authentication device 5 comes only to the database 2 or its upstream voice dialogue system 2a. In other words, the data flows 6e and 6f from the database 2 or its upstream voice dialogue system 2a via the switching node 4 to the user telephone 1 and thus to the caller no longer contain any signals from the authentication device 5 and can therefore not be used abusively. This data is therefore shown in Fig. 2 by a white arrow 9, in which no more authentication signals are included.

Thus, the system can be operated as a three-way conference, are transmitted in the critical data only from a first participant to a fixed predetermined second participant. Thus, the answers of the caller to prompts of the speech dialogue system ("Talk to me ...") immediately recorded by the authentication device 5 and analyzed. The communication between the authentication device and the speech dialogue system via DTMF and is not heard by the caller, as these sound sequences are filtered out.

Via an application programming interface (API), the speech dialogue system 2a of the database 2 is notified via a positive authentication. The application does not require any data connection to the authentication system 5 in the embodiment according to the invention.

In addition to the communication via DTMF sequences, an evaluation of the speech (speech recognition) in the authentication device 5 is also possible. In this case, no adjustments to the authentication system are necessary. Speech recognition in this case has a 100% hit rate because the text sequences are hard-coded in the authentication system. An announcement about a successful authentication can be recognized by the remote authentication.

The invention thus has the advantage that the communication between an authentication system and a speech dialogue system can be realized without having to use any other than the telephone interface. The authentication system 5 can be located here at any location with network connection.

In the description so far, it has been assumed that the separation of the authentication signals and the signals intended for the user takes place in the switching node 4. However, it will be apparent to those skilled in the art that this is only an example, and that this separation of the signals is just as well in the actual access control, i. an ISDN card or the interactive voice dialogue system 2a can take place.

Furthermore, the method according to the invention is not limited to an authentication attempt, but several attempts can be permitted, in which then an averaging of the individual results of the authentication can also take place if appropriate. In other words, a voice sample is taken comprising at least one sequence with several syllables or words, ie the user is asked several times (by default or depending on the previous authentication results) to utter one word and / or syllable, one each Voice sample is recorded. The individual voice samples are each compared to a comparison data set, whereby an overall hit rate is determined, ie a quotient of the successful tests and the total number of attempts. The release or denial of access to the desired information by the user then success as a function of the determined hit rate by the access control 2a.

In order to avoid as far as possible failures due to line disturbances in this connection, it is preferable to use voice filtering (not shown) for eliminating disturbances of the telephone line.

reference numeral

1

User telephone, 1a user

2

Database, 2a access control, esp. Interactive language system

3

telecommunications network

4

switching node

5

authentication device

6

Access line, 6a - 6f data flows

7

Identification signal from user

8th

Comparison result of authentication device

9

Authentication result of database access control

Claims (11)

1. Method for authorizing access to a database (2) by a user in a telecommunication network (3), the telecommunication network (3) comprising a user interface (1), an access control (2a) for the database (2) and an authentification device (5), connected with each other through a switching node (4) in the telecommunication network (3), the method comprising the following steps: the access control (2a) makes the user give a voice specimen,
   the authentification device (5) compares the voice specimen with a comparative data record assigned to the user, which had been stored beforehand in a storage,
   the access control (2a) grants or denies the user access to the desired information as a function of the result of the comparison,
   the authentification device (5) outputting the result of the comparison through the switching node (4) to the access control (2a),
   characterized in that the access control (2a) informs the user of the result of the comparison by means of the switching node (4), signals being masked out by the authentification device (5).
2. Method according to claim 1, characterized in that the telecommunication network (3) is an ISDN network.
3. Method according to claim 2, characterized in that the B-channels are interconnected in such a way that in the switching node (4), in each case only that receiving time slot for signals from the authentification device (5) is mixed with a sending time slot for the user telephone (1), in which no authentification signals were received.
4. Method according to any of the preceding claims, characterized in that the result of the comparison is transmitted between the authentification device (5) and the access control (2a) in the form of DTMF sequences.
5. Method according to claim 1 or 2, characterized in that the voice specimen comprises at least one sequence of several syllables or words, the user being asked several times to pronounce a word and/or a syllable, and each time a voice specimen being recorded,
   each of the several voice specimens being compared with a comparative data record, determining a hit rate, and
   the grant or denial of the user's access to the desired information being effected as a function of the hit rate determined by the access control (2a).
6. Method according to any of the preceding claims, in which a voice filtering device is used to eliminate interferences of the telephone line.
7. Method according to any of the preceding claims, in which the masking out of signals by the authentification device (5) from the signals of the access control (2a) for the user (1 a) is effected in the switching node.
8. Method according to any of the preceding claims, in which the access control (2a) is an interactive voice dialog system.
9. Switching node (4) for switching signals in a telecommunication network (3) between a user telephone (1), an authentification device (5) and a database (2) for authorizing an access to a database (2) through a user in the telecommunication network (3), according to any of claims 1 to 8,
   those signals which the authentification device (5) sends to the access control (2a) of the database (2) being masked out from the signals sent by the access control (2a) of the database (2) to the user telephone (1).
10. Switching node (4) according to claim 7, characterized in that it is an ISDN switching node in which the B-channels are interconnected in such a way that in each case only that receiving time slot for data from the authentification device (5) is mixed with a sending time slot for data for the user telephone (1), in which no authentification signals were received.
11. Switching node (4) according to claim 7 or 8, characterized in that it comprises a voice filtering device for eliminating interferences of the telephone line.

Images